Power supply simulator. / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Power supply simulator.

1 2 3 4 next last

Move Thread

LAN_403

Ash Small

Tue May 01 2018, 03:48PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

I've been using this power suply simulator to design an RC smoothing filter for an audio amp supply.

It's downloadable from the Duncan Amps website, and it's pretty simple to use, once you change the settings to RC filter from C filter.

I had what started out as a 5 amp 24 volt battery charger, and ripped the regulator circuit out, just keeping the transformer, rectifier and 4700uF smoothing cap. With a 5 amp load I was seeing about 2.5V ripple.

I tried playing around with the simulator to get a feel for it, after measuring the DC resistance of the transformer secondary (0.6 Ohms), and noticed that things began to get interesting if I doubled the capacitance in successive stages, with 0.34 Ohms between each stage.

Peak voltage was around 31.8V, which I pretty much simulated, and using the original 4700uF cap as the first stage, and doubling up on each successive stage requires another 30 caps, which, courtesy of Avalanche, should me heading my way later today.

It's interesting playing with the simulator, seeing what effect changing the load, etc has on voltage and ripple. The values I'm playing with at the moment are giving between one and two millivolts ripple, and voltage fluctuates a few volts with a varying load.

As it's for a Class A design, load should be pretty constant in use.

I intend to use additional RC filters on the pre amp stages, particularly the input stage, which I've tested at 18V rail to rail, so I can afford to lose a few more volts in additional filtering.

The output stage is still in the experimental stage, it will evolve around a handfull of those new SiC power JFET's.

Sulaiman

Wed May 02 2018, 07:25AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

Using Duncan's amp simulator,
choose to add an inductive filter
- L.C filters reduce output voltage ripple,
and significantly reduce the twice-per-cycle current spikes,
which contribute to rfi.

Ash Small

Wed May 02 2018, 11:29AM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

That's a very good point, Sulaiman.

I've been concentrating on RC circuits anp impedence recently. although I'm looking at using an LC network for filtering for a reverb stage.

Any tips on inductors?

I'm guessing winding some on iron powder cores is the obvious solution?

I have one I wound a few years ago that should take the current, I'll try and measure the inductance.

The more voltage I have across the power stage, the higher the load impedence can be, requiring less output capacitance for the same low frequency response.

Sulaiman

Wed May 02 2018, 04:10PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

Laminated steel cores with an air gap would be cheapest and most compact,
even a little series inductance reduces the current spikes enormously,
e.g. dc filter chokes/inductors from old equipment, often yellow and/or green.(powdered iron)
NOT high permeability ring/toroid cores,
ferrite with an airgap is ok but steel is cheaper.

The only real problem with inductive smoothing are;
. size, weight and cost
. the output voltage droops as output current is drawn
ok for pmpo-rated amplifiers, or if a supply voltage regulator is used.
. depending upon capacitance and inductance values,
when a large current output is suddenly stopped (e.g. end of a loud section of music)
the power supply voltage can 'bounce' above nominal for a while.

Putting a capacitor across each of the diodes in a rectifier bridge (e.g. 100nf ceramic)
reduces rfi generated when the diodes switch on/off each 1/2-cycle of the mains,

not a high level interference signal but quite difficult to filter out.
Because the switching causes rf bursts at 2x mains frequency,
we hear the rfi interference as hum, with a rough edge (harmonics)

As I'm sure you've read, eliminating hum has many approaches;
. single-point earthing, impractical in most cases .but the principles are good
. ferrite beads/rings on ALL cables, power, signal in, signal out, loudspeaker cables ....
because long lengths of wire/cable act as receiving antennas for external rfi
and internally they behave as rfi transmiter antennas :).

Ash Small

Wed May 02 2018, 06:34PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

I have a couple of T225-26 cores here, the yellow/white ones (57x35x25), and can get some T300-26 (77x49x25, I think). I'm planning on running some tests this evening. One is fully wound with 14SWG, the other I've just wound ten turns onto. I have some known value capacitors and a sig gen and scope, so can work out resonant frequency, then work out inductance, but inductance with chokes, from what I remember, is current dependent. I can see some trial and error involved, it should keep me busy for a while. If that doesn't work, I'll have to cannibalise some laminated core transformers.

Current should be constant in a single ended Class A design, regardless of signal strength. It ideally wants to be fed from a constant current supply.

I'm guessing PP caps will work at least as aell as ceramic for snubbing the rectifier. I think I have some 100nF, I certainly have 0.1uF tropical fish. I don't like using ceramics, it seems to make more sense to use PP or PET.

I've not found my stash of ferrite rings since moving house, but I'm sure they'll turn up somewhere

The simulator gives less than 200uV ripple with four 2mH chokes @ 5A. I'll see what figures I come up with for these T225-26 cores.

Ash Small

Wed May 02 2018, 08:54PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

NEW INFORMATION;-

So the T225-26 core (these come in different widths, and different manufacturers use different designations, but this one is 25mm thick), when connected to a 10uF PP capacitor, has a resonant frequency of 1.3kHz, which the online calculators tell me equates to 1.5mH, right in the ballpark.

With ten turns on one I get a resonant frequency, when connected to the same cap, of 13kHz, which the calculators tell me equates to 0.015mH.

I'm thinking if I order 4xT300-26 (25mm thick version) (77mm x 49mm x25, or thereabouts), I can easily wind 2mH or more, probably using PVC coated wire of 17A rating or more.

I think it's worth doing just for the increased voltage it gives me (5 volts more across the load than using resistors, and I expect to lose around ten of those biasing the JFET's).

I'm even considering extra chokes, similar to those you'd find in a parafeed circuit, I'm thinking it could only improve things further.

EDIT:- This place selis small quantities of cores at very reasonable prices, with what seems to be an Â£11 flat rate next day delivery charge.

DerAlbi

Wed May 02 2018, 11:46PM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

At some point its easier so use a simple linear voltage regulator. They suppress low frequency ripple up to several kHz pretty well, if you pick the right one

Ash Small

Thu May 03 2018, 05:53PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

DerAlbi wrote ...

At some point its easier so use a simple linear voltage regulator. They suppress low frequency ripple up to several kHz pretty well, if you pick the right one

I did look at voltage regulators, and zeners, to control the voltage, but apart from the voltage drops involved, any device using silicon for switching creates noise.

I've recently been studying RC circuits and impedence (not that the subject is difficult to understand, especially if you understand LC circuits, but because it wasn't something I felt I 'needed' to understand until recently, so I'd not bothered), and I'd asked for advice relating to people's experiences using zeners and regulators, as well as any advice/links on RC networks, and was pointed towards the Duncan Amps simulator, and realised I could get similar or better regulation to zeners or regulators using resistors/capacitors for a similar voltage drop.

It didn't occur to me to consider inductors until Sulaiman suggested it further up this thread (thanks for pointing out the obvious, Sulaiman

). The capacitors were already on their way by this time, but I realised immediately that an LC network made a lot of sense, and put the numbers into the simulator, using the same cap values, and realised that with just a couple of mH between cap banks the regulation was 10X better than resistors, and I had an extra 5V at the output.

Also, chokes are useful for current regulation in a single ended Class A design, I'm now thinking of adding even more chokes, and have ordered extra T300-26 toroids to experiment with.

I'd sooner spend money on powdered iron toroids for a much improved power supply than spend money on large heatsinks to dissipate wasted power in a (possibly) unreliable and noisy supply

Sulaiman

Fri May 04 2018, 05:38AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

In the days before neodymium or even ceramic magnetic materials,
AlNiCo magnets were available but expensive,
so often the loudspeaker had a large coil of wire as an electromagnet for the fixed magnetic field,
by putting the loudspeaker magnetising coil in the +HT line,
the loudspeaker magnet was energised,
and the loudspeaker magnet inductance 'smoothed' the +HT.
======================================
The powdered iron cores sound expensive,
I would make an E-core and I-core from the laminations of a steel transformer,
and use those to make the inductor(s) with an air gap between E and I.
========================================
A common method of reducing hum was to construct the amplifier,
then move and/or rotate transformers and inductors,
one source of hum cancelling another.
The reduction of hum can be significant.
===================================== ===
-26 material is good for dc chokes because the iron powder core has in effect a distributed air gap,
i.e. it leaks equally in all directions

Ash Small

Fri May 04 2018, 05:23PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

The T300-26 cores arrived this morning, very reasonably priced from the link above, Â£6 something each, plus Â£11 next day delivery. Capacitors arrived yesterday.

I'm assuming 2.5mm^2 will be ok to use, the seven strand PVC covered stuff is about 3mm diameter, so approx 120 turns per layer, from initial estimates.

They have more CSA than the 225 cores that are 25mm thick, and twice the window area. The fully wound 225 pictured below is 1.5mH inductance with 2mm OD enamelled wire (SWG 14, I think), so I should be able to get 2mH from the T300.

I'm just hoping it doesn't saturate with 5A current.

1 2 3 4 next last

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob